Apparatus and method for coding and decoding image signals

Abstract

A compression-coded high definition television signal is decoded and converted to a standard NTSC signal without loss of interlacing so that the resulting picture exhibits smooth motion. The compression-coding of the HDTV signal includes DCT processing to produce an 8×8 block of coefficient data. Upon decoding, the upper left quadrant of the coefficient data array is used to form a 4×4 subblock of coefficient data but the last row of elements of the subblock is replaced by corresponding elements of the eighth row of the 8×8 block of coefficient data.

What is claimed is: 
     
       1. A method of decoding image data in the form of a block of coefficient data elements produced by performing an orthogonal transformation upon a block of picture element data, comprising the steps of: forming from said block of coefficient data elements a first subblock having a smaller range than said block;   forming a modified subblock by replacing at least one element of said first subblock with a corresponding at least one element of said block from outside of said first subblock; and   performing an inverse orthogonal transformation on said modified subblock to produce reconstituted picture element data.   
     
     
       2. A method of decoding image data according to claim 1, wherein said block of coefficient data elements represents coefficient data elements arrayed in m rows and n columns, and said first subblock represents an intersection of a first j rows and a first k columns of said block of coefficient data elements, j being less than m, k being less than n; and said step of forming said modified subblock includes replacing a last j row of elements of said first subblock with corresponding elements of a last m row of said block, j, k, m, and n being positive integers. 
     
     
       3. A method of decoding image data according to claim 1, wherein said orthogonal transformation is a discrete cosine transform and said inverse orthogonal transformation is an inverse discrete cosine transform. 
     
     
       4. A method of decoding image data according to claim 1, wherein said block of picture element data is produced from a frame of picture element data consisting of two interlaced fields, said coefficient data elements having been produced on the basis of differences between reference picture data and encoding predictive picture data generated with motion compensation in one of a frame predictive mode and a field predictive mode; and further comprising the steps of: generating decoding predictive picture data using motion compensation in said one of said predictive modes; and   combining said reconstituted picture element data and said decoding predictive picture data.   
     
     
       5. A method of decoding image data according to claim 4, wherein said decoding predictive picture data includes picture elements that correspond to picture elements of said encoding predictive picture data. 
     
     
       6. A method of decoding image data according to claim 4, wherein positions of picture elements of said decoding predictive picture data coincide with positions of corresponding picture elements of said encoding predictive picture data, if said one of said predictive modes is said field predictive mode. 
     
     
       7. A method of decoding image data according to claim 4, wherein a field parity of picture elements of said decoding predictive picture data coincides with a field parity of corresponding picture elements of said encoding predictive picture data, if said one of said predictive modes is said frame predictive mode. 
     
     
       8. A method of decoding image data according to claim 7, wherein picture elements of said decoding predictive picture data are formed by interpolation if corresponding picture elements of said encoding predictive picture data were formed by interpolation. 
     
     
       9. A method of decoding image data in the form of a block of coefficient data elements produced by performing an orthogonal transformation upon a block of picture element data, comprising the steps of: forming from said block of coefficient data elements a first subblock having a smaller range than said block;   forming a modified subblock by replacing at least one element of said first subblock with a sum of said at least one element and a corresponding at least one element of said block from outside of said first subblock; and   performing an inverse orthogonal transformation on said modified subblock to produce reconstituted picture element data.   
     
     
       10. A method of decoding image data according to claim 9, wherein said orthogonal transformation is a discrete cosine transform and said inverse orthogonal transformation is an inverse discrete cosine transform. 
     
     
       11. A method of decoding image data according to claim 9, wherein said block of picture element data is produced from a frame of picture element data consisting of two interlaced fields, said coefficient data elements having been produced on the basis of differences between reference picture data and encoding predictive picture data generated with motion compensation in one of a frame predictive mode and a field predictive mode; and further comprising the steps of: generating decoding predictive picture data using motion compensation in said one of said predictive modes; and   combining said reconstituted picture element data and said decoding predictive picture data.   
     
     
       12. A method of decoding image data according to claim 11, wherein said decoding predictive picture data includes picture elements that correspond to picture elements of said encoding predictive picture data. 
     
     
       13. A method of decoding image data according to claim 11, wherein positions of picture elements of said decoding predictive picture data coincide with positions of corresponding picture elements of said encoding predictive picture data, if said one of said predictive modes is said field predictive mode. 
     
     
       14. A method of decoding image data according to claim 11, wherein a field parity of picture elements of said decoding predictive picture data coincides with a field parity of corresponding picture elements of said encoding predictive picture data, if said one of said predictive modes is said frame predictive mode. 
     
     
       15. A method of decoding image data according to claim 14, wherein picture elements of said decoding predictive picture data are formed by interpolation if corresponding picture elements of said encoding predictive picture data were formed by interpolation. 
     
     
       16. A method of decoding image data in the form of a block of coefficient data elements produced by performing an orthogonal transformation upon a block of picture element data, wherein said block of coefficient data represents coefficient data elements arrayed in m rows and n columns, comprising the steps of: forming from said block of coefficient data elements a first subblock having a smaller range than said block, said first subblock represents an intersection of the first j rows and first k columns of said block of coefficient data elements, j being less than m, k being less than n;   forming a modified subblock by replacing at least each element of the last row of said first subblock with a sum of said element and a corresponding element of the last row of said block; and   performing an inverse orthogonal transformation on said modified subblock to produce reconstituted picture element data.   
     
     
       17. A method of decoding image data according to claim 16, wherein m=n and j=k. 
     
     
       18. A method of decoding image data according to claim 17, wherein m=n=8 and j=k=4. 
     
     
       19. An apparatus for decoding a block of image data that was generated by performing an orthogonal transformation upon a block of picture element data from a frame of picture elements consisting of two interlaced fields, said orthogonal transformation producing a block of coefficient data elements representing coefficient data elements arrayed in m rows and n columns, and then quantizing and variable-length-encoding said block of coefficient data elements; the apparatus comprising: variable length decoding means for performing variable length decoding upon said image data to form decoded data;   dequantization means for dequantizing the decoded data to recover said block of coefficient data elements;   subblock forming means for forming a j row by k column subblock from said recovered block of coefficient data elements, j being less than m, k being less than n, the first j-1 rows of said subblock being formed as an intersection of the first j-1 rows and first k columns of said recovered block of coefficient data elements, each element of the jth row of said subblock being formed as the sum of respective corresponding elements of the jth and mth rows of said recovered block of coefficient data elements; and   inverse orthogonal transformation means for performing an inverse orthogonal transformation upon said subblock.   
     
     
       20. An apparatus according to claim 19, wherein m=n and j=k, with j, k, m, and n being positive integers. 
     
     
       21. An apparatus according to claim 19, wherein said orthogonal transformation is a discrete cosine transform and said inverse orthogonal transformation is an inverse discrete cosine transform.